The worst case is that it could require more space due to a collation that forces everything to a Unicode comparison based on the data types, but it should not be a tremendous amount of difference. If you are making comparisons between Unicode-Unicode values, you'll need the same space.